Can't use "volume.shared" in Compontent Chart

[johnnypea]

error deploying: error deploying php: deployments[php].helm.componentChart: component values are incorrect: yaml: unmarshal errors:
  line 18: field shared not found in type latest.VolumeMountVolumeConfig

https://devspace.sh/component-chart/docs/configuration/containers#volumeshared

volumeMounts:
- containerPath: /var/run/php
  volume:
    name: php-socket
    readOnly: false
    shared: true

[FabianKramm]

@johnnypea thanks for creating this issue! Yes seems like this is not part of the latest spec in devspace, we'll fix that for the next release!

[johnnypea]

@FabianKramm thanks!

Please, could you give me working example how can I use this sharing a volume between the pods. It is not very clear from the docs. Should "shared" be set for every pod which is using that shared volume?

[LukasGentele]

@johnnypea As far as I see it in the chart's templates, the shared: true option creates a PersistentVolumeClaim separately with name php-stocket as in your example. That means multiple deployments can mount this PVC. If you have shared: false, instead of creating a PVC directly, the chart will create a PVC Template as part of the StatefulSet which makes it very hard to mount it from 2 different deployments. Details in this commit: loft-sh/component-chart@d8c5e1a

EDIT: Yes, shared: true should be set for every deployment.

[johnnypea]

@LukasGentele @FabianKramm I finally got to test it on 5.18.3 but still no luck getting it work...

00:05:32 [fatal]  error deploying: error deploying pwa: Unable to deploy helm chart: error during 'helm upgrade pwa --namespace dev-intraspace --values /var/folders/mc/09gm4cz13bxgpj5nb8nkg_380000gn/T/348670964 --install component-chart --repo https://charts.devspace.sh --repository-config='' --version 0.8.4 --kube-context minikube': Release "pwa" does not exist. Installing it now.
Error: rendered manifests contain a resource that already exists. Unable to continue with install: PersistentVolumeClaim "pwa-dist" in namespace "dev-intraspace" exists and cannot be imported into the current release: invalid ownership metadata; annotation validation error: key "meta.helm.sh/release-name" must equal "pwa": current value is "app"
 => exit status 1

deployments:
- name: app
  # This deployment uses `helm` but you can also define `kubectl` deployments or kustomizations
  helm:
    # We are deploying the so-called Component Chart: https://devspace.sh/component-chart/docs
    componentChart: true
    # Under `values` we can define the values for this Helm chart used during `helm install/upgrade`
    # You may also use `valuesFiles` to load values from files, e.g. valuesFiles: ["values.yaml"]
    values:
      containers:
      - name: caddy-container
        image: caddy
        volumeMounts:
        - containerPath: /srv/pwa/dist
          volume:
            name: pwa-dist
            readOnly: true
            shared: true
      volumes:
      - name: pwa-dist
        size: 5Gi
- name: pwa
  helm:
    componentChart: true
    values:
      containers:
      - image: pwa-prod
        volumeMounts:
        - containerPath: /usr/src/pwa/dist
          volume:
            name: pwa-dist
            shared: true
      volumes:
      - name: pwa-dist
        size: 5Gi

When I removed the volume definition I just got:

00:10:10 [fatal]  error deploying: error deploying pwa: Unable to deploy helm chart: error during 'helm upgrade pwa --namespace dev-intraspace --values /var/folders/mc/09gm4cz13bxgpj5nb8nkg_380000gn/T/2846785660 --install component-chart --repo https://charts.devspace.sh --repository-config='' --version 0.8.4 --kube-context minikube': Release "pwa" does not exist. Installing it now.
Error: Deployment.apps "pwa" is invalid: spec.template.spec.containers[0].volumeMounts[0].name: Not found: "pwa-dist"
 => exit status 1

[FabianKramm]

@johnnypea when you get the resource already exist error, can you do a devspace purge and then devspace dev again?

[johnnypea]

@FabianKramm I did devspace purge --all and kubectl delete persistentvolumeclaims --all but still the same result. I tried deploying in new namespace but it didn't help.

This pod yaml if it would be any help. I can't see "shared" definition there, is that correct?

apiVersion: v1
kind: Pod
metadata:
  annotations:
    helm.sh/chart: component-chart-0.8.4
  creationTimestamp: "2022-02-03T09:42:52Z"
  generateName: app-
  labels:
    app.kubernetes.io/component: app
    app.kubernetes.io/managed-by: Helm
    app.kubernetes.io/name: devspace-app
    controller-revision-hash: app-596fd7dd97
    statefulset.kubernetes.io/pod-name: app-0
  name: app-0
  namespace: dev-intraspace
  ownerReferences:
  - apiVersion: apps/v1
    blockOwnerDeletion: true
    controller: true
    kind: StatefulSet
    name: app
    uid: 0c90d19c-776b-47ff-971e-776a7ff55ccb
  resourceVersion: "220929"
  uid: a7037f7e-2fa4-41c0-ab90-5378420efdd0
spec:
  containers:
  - env:
    - name: DATABASE_URL
      value: postgresql://symfony:ChangeMe@database:5432/app?serverVersion=13
    - name: MERCURE_JWT_SECRET
      value: '!ChangeMe!'
    - name: MERCURE_PUBLIC_URL
      value: https://localhost/.well-known/mercure
    - name: MERCURE_URL
      value: http://caddy/.well-known/mercure
    image: php:aAflOfS
    imagePullPolicy: IfNotPresent
    name: php-container
    resources: {}
    terminationMessagePath: /dev/termination-log
    terminationMessagePolicy: File
    volumeMounts:
    - mountPath: /var/run/php
      name: php-socket
    - mountPath: /usr/local/etc/php/conf.d/symfony.ini
      name: phpdevini-config
      readOnly: true
      subPath: symfony.ini
    - mountPath: /var/run/secrets/kubernetes.io/serviceaccount
      name: kube-api-access-8vrbr
      readOnly: true
  - env:
    - name: MERCURE_PUBLISHER_JWT_KEY
      value: '!ChangeMe!'
    - name: MERCURE_SUBSCRIBER_JWT_KEY
      value: '!ChangeMe!'
    - name: SERVER_NAME
      value: localhost:8080, localhost:4443, caddy:80
    image: caddy:HdVoAwM
    imagePullPolicy: IfNotPresent
    name: caddy-container
    resources: {}
    terminationMessagePath: /dev/termination-log
    terminationMessagePolicy: File
    volumeMounts:
    - mountPath: /var/run/php
      name: php-socket
      readOnly: true
    - mountPath: /srv/pwa/dist
      name: pwa-dist
      readOnly: true
    - mountPath: /data
      name: caddy-data
    - mountPath: /config
      name: caddy-config
    - mountPath: /etc/caddy/Caddyfile
      name: caddyfile-config
      readOnly: true
      subPath: Caddyfile
    - mountPath: /var/run/secrets/kubernetes.io/serviceaccount
      name: kube-api-access-8vrbr
      readOnly: true
  dnsPolicy: ClusterFirst
  enableServiceLinks: true
  hostname: app-0
  imagePullSecrets:
  - name: devspace-auth-docker
  nodeName: minikube
  preemptionPolicy: PreemptLowerPriority
  priority: 0
  restartPolicy: Always
  schedulerName: default-scheduler
  securityContext: {}
  serviceAccount: default
  serviceAccountName: default
  subdomain: app-headless
  terminationGracePeriodSeconds: 5
  tolerations:
  - effect: NoExecute
    key: node.kubernetes.io/not-ready
    operator: Exists
    tolerationSeconds: 300
  - effect: NoExecute
    key: node.kubernetes.io/unreachable
    operator: Exists
    tolerationSeconds: 300
  volumes:
  - name: php-socket
    persistentVolumeClaim:
      claimName: php-socket-app-0
  - name: caddy-data
    persistentVolumeClaim:
      claimName: caddy-data-app-0
  - name: caddy-config
    persistentVolumeClaim:
      claimName: caddy-config-app-0
  - configMap:
      defaultMode: 420
      name: caddyfile-config
    name: caddyfile-config
  - configMap:
      defaultMode: 420
      name: phpdevini-config
    name: phpdevini-config
  - name: pwa-dist
    persistentVolumeClaim:
      claimName: pwa-dist
  - name: kube-api-access-8vrbr
    projected:
      defaultMode: 420
      sources:
      - serviceAccountToken:
          expirationSeconds: 3607
          path: token
      - configMap:
          items:
          - key: ca.crt
            path: ca.crt
          name: kube-root-ca.crt
      - downwardAPI:
          items:
          - fieldRef:
              apiVersion: v1
              fieldPath: metadata.namespace
            path: namespace
status:
  conditions:
  - lastProbeTime: null
    lastTransitionTime: "2022-02-03T09:42:53Z"
    status: "True"
    type: Initialized
  - lastProbeTime: null
    lastTransitionTime: "2022-02-03T09:42:55Z"
    status: "True"
    type: Ready
  - lastProbeTime: null
    lastTransitionTime: "2022-02-03T09:42:55Z"
    status: "True"
    type: ContainersReady
  - lastProbeTime: null
    lastTransitionTime: "2022-02-03T09:42:53Z"
    status: "True"
    type: PodScheduled
  containerStatuses:
  - containerID: docker://db6c1c3976ac622e083a2707f4f04b332eb1280eeb30f271704ab77d8231bde5
    image: caddy:EOwtPGO
    imageID: docker://sha256:42f108cc22d108667641908aad865e3050f23815e69c662340f44739752df607
    lastState: {}
    name: caddy-container
    ready: true
    restartCount: 0
    started: true
    state:
      running:
        startedAt: "2022-02-03T09:42:55Z"
  - containerID: docker://42de142e91e5445511d1ef1b45cbe8dbb88eea2c913207483f1699bae5a2b69d
    image: php:aAflOfS
    imageID: docker://sha256:e9838b262e9c90f3662a1688f4ea7dd35a2bd5b76a82462d402a61a60e129ab8
    lastState: {}
    name: php-container
    ready: true
    restartCount: 0
    started: true
    state:
      running:
        startedAt: "2022-02-03T09:42:55Z"
  hostIP: 192.168.49.2
  phase: Running
  podIP: 172.17.0.3
  podIPs:
  - ip: 172.17.0.3
  qosClass: BestEffort
  startTime: "2022-02-03T09:42:53Z"

[FabianKramm]

@johnnypea thanks for the info, can you share your devspace.yaml that fails during devspace dev? Do you have multiple deployments that somehow reference the same pvc?

[johnnypea]

@LukasGentele Yes, I do. I sent it in my previous comment. Do you need any more details?

[FabianKramm]

@johnnypea thanks for the info, I guess thats currently a limitation of the component chart as it tries to create 2 PVC's here. You are probably better off to create your own custom helm chart that does this the way you want with a single PVC and multiple deployments

[johnnypea]

@FabianKramm would it be possible to use it like this?

deployments:
- name: pwa-dist-pvc
  kubectl:
    manifests:
      - pwa/deploy/pwa-dist-pvc.yaml
- name: pwa
  helm:
    componentChart: true
    values:
      containers:
      - image: pwa-prod
        volumeMounts:
        - containerPath: /usr/src/pwa/dist
          volume:
            name: pwa-dist

pwa/deploy/pwa-dist-pvc.yaml

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: pwa-dist
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 5Gi

Getting this error:

Error: Deployment.apps "pwa" is invalid: spec.template.spec.containers[0].volumeMounts[0].name: Not found: "pwa-dist"

[FabianKramm]

@johnnypea no, the problem is that the pvc already exists and you need to separate the deployments, where 1 deployment wouldn't create the persistent volume claim